Researcher, Post Training

Lovable · Coding AI · Stockholm, Sweden · Engineering

Lovable is seeking an engineer with experience in post-training large language models at scale to own their full post-training pipeline. This role involves translating research into production training recipes for code generation and agent workloads, with a focus on shipping improved models quickly. The position requires strong production code skills, familiarity with ML frameworks like PyTorch or JAX, and an understanding of preference optimization, reward modeling, and alignment techniques. The engineer will also build evaluation systems and operate production systems for training jobs.

What you'd actually do

Own the full lifecycle of Lovable's post-training pipeline - from data curation and training runs through evaluation and deployment
Apply and adapt reinforcement learning, preference optimization, and supervised fine-tuning methods to make our models better at generating code, reasoning about user intent, and acting as reliable agents
Build the evaluation and experimentation infrastructure that tells us whether a model change actually helps users - covering helpfulness, safety, latency, and reliability
Develop and operate the production systems that run training jobs at scale, including GPU orchestration and data pipelines
Work across team boundaries with our agent, product, and infrastructure engineers to turn model gains into product improvements users can feel

Skills

Required

post-training jobs on large language models (RFT/RLVR, preference optimization, or similar)
production code
ML framework (PyTorch, JAX)
distributed training setups
GPU clusters
math behind preference optimization, reward modeling, and alignment techniques
evaluation systems
trace model quality regression
ship models

Nice to have

worked on code generation or agentic use cases
put post-trained models into the hands of real users
owned the full loop: curating data, running training, evaluating results, deploying, and monitoring in production
habit of reading a paper on Monday and having a prototype running by Friday
experimented with speculative decoding or similar techniques
strong views on evaluation methodology
built evals that predict user satisfaction
published or contributed meaningfully to the open-source ML ecosystem

What the JD emphasized

post-training at scale
production training recipes
code generation and agent workloads
research into production within days or weeks
production infrastructure
ship
large language models
production code
distributed training setups
GPU clusters
preference optimization
reward modeling
alignment techniques
evaluation systems
model quality regression
code generation
agentic use cases
data curation
training runs
evaluation
deployment
monitoring in production
speculative decoding
evaluation methodology
open-source ML ecosystem
lifecycle of post-training pipeline
reinforcement learning
supervised fine-tuning
generating code
reasoning about user intent
reliable agents
evaluation and experimentation infrastructure
helpfulness
safety
latency
reliability
production systems
GPU orchestration
data pipelines
agent engineers
product engineers
infrastructure engineers
model gains into product improvements
investigate and resolve failures
training recipe
data issue
serving regression
move fast

Other signals

post-training at scale
production training recipes
code generation and agent workloads
research into production within days or weeks
production infrastructure
ship
large language models
production code
distributed training setups
GPU clusters
preference optimization
reward modeling
alignment techniques
evaluation systems
model quality regression
code generation
agentic use cases
data curation
training runs
evaluation
deployment
monitoring in production
speculative decoding
evaluation methodology
open-source ML ecosystem
lifecycle of post-training pipeline
reinforcement learning
supervised fine-tuning
generating code
reasoning about user intent
reliable agents
evaluation and experimentation infrastructure
helpfulness
safety
latency
reliability
production systems
GPU orchestration
data pipelines
agent engineers
product engineers
infrastructure engineers
model gains into product improvements
investigate and resolve failures
training recipe
data issue
serving regression
move fast

Apply on company site

● Active

Posted 3mo ago · 80 days open

AI score: 9/10
Stage: Post-train Agent
Location: Stockholm, Sweden
Role: Senior · Builder
Function: Engineering
Domain: general
Maturity: Scaling

Skills

Agents & Autonomy

Agent OrchestrationAgentic SystemsPlanning & Reasoning

Data Engineering

Data LabelingData Pipelines

General Experience & Skills

Customer-Facing ExperienceOpen Source ContributionsPerformance OptimizationProduction-Grade CodeSystem Design

Infrastructure & Systems

CI/CDGPU ComputingModel Serving

LLM & Foundation Models

AI SafetyLLM Evaluation & GradingModel CompressionSpeculative Decoding

ML Ops & Evaluation

A/B TestingModel Monitoring

ML Techniques

Code GenerationDirect Preference Optimization (DPO)Model Post-TrainingReinforcement Learning (RL)Supervised Fine-Tuning (SFT)

Research & Credentials

ML Research

Read full job description

**TL;DR: ** Lovable lets over 2 million people build software using plain language, and the models behind it need to be exceptional. We're hiring an engineer who has gotten their hands dirty with post-training at scale and wants to do it again for one of the fastest-growing AI products in the world.

You'll own our full post-training pipeline: translating the latest research into production training recipes, adapting them for code generation and agent workloads, and putting improved models in front of users fast. The goal is to get promising research into production within days or weeks, not months. This isn't an academic research position - you'll spend as much time in production infrastructure as in training configs, and your success is measured by what ships.

Why Lovable?

Lovable lets anyone and everyone build software with any language. From solopreneurs to Fortune 100 teams, millions of people use Lovable to transform raw ideas into real products - fast. We are at the forefront of a foundational shift in software creation, which means you have an unprecedented opportunity to change the way the digital world works. Over 2 million people in 200+ countries already use Lovable to launch businesses, automate work, and bring their ideas to life. And we’re just getting started.

We’re a small, talent-dense team building a generation-defining company from Stockholm. We value extreme ownership, high velocity, and low-ego collaboration. We seek out people who care deeply, ship fast, and are eager to make a dent in the world.

What we’re looking for

You've personally run post-training jobs on large language models - RFT/RLVR, preference optimization, or similar. Not just called APIs or written prompts, but actually trained and iterated on models
You can write solid production code. The systems you build need to run reliably, not just produce interesting research artifacts
You're fluent in at least one major ML framework (PyTorch, JAX) and comfortable working with distributed training setups and GPU clusters
You understand the math behind preference optimization, reward modeling, and alignment techniques - and can reason about when each approach fits
You've built or significantly contributed to evaluation systems that capture real-world quality, not just benchmark scores
You can trace a model quality regression from user-facing symptoms back through serving, inference, and training - and you enjoy doing it
You want to ship. Research taste matters, but at Lovable the question is always "how fast can we get this to users?"

Preferred:

You've worked on code generation or agentic use cases specifically
You've put post-trained models into the hands of real users and seen how they hold up at scale
You've owned the full loop: curating data, running training, evaluating results, deploying, and monitoring in production
You have a habit of reading a paper on Monday and having a prototype running by Friday
You've experimented with speculative decoding or similar techniques to improve model efficiency
You have strong views on evaluation methodology and have built evals that actually predict user satisfaction
You've published or contributed meaningfully to the open-source ML ecosystem

What you’ll do

Own the full lifecycle of Lovable's post-training pipeline - from data curation and training runs through evaluation and deployment
Apply and adapt reinforcement learning, preference optimization, and supervised fine-tuning methods to make our models better at generating code, reasoning about user intent, and acting as reliable agents
Build the evaluation and experimentation infrastructure that tells us whether a model change actually helps users - covering helpfulness, safety, latency, and reliability
Develop and operate the production systems that run training jobs at scale, including GPU orchestration and data pipelines
Work across team boundaries with our agent, product, and infrastructure engineers to turn model gains into product improvements users can feel
Investigate and resolve failures end-to-end - whether the root cause is in a training recipe, a data issue, or a serving regression
Read papers, run experiments, and move fast: the goal is to get promising research into production within days or weeks, not months

About your application

Please submit your application in English. It’s our company language, so you’ll be speaking lots of it if you join.

We treat all candidates equally - if you’re interested, please apply through our careers portal.